Electric vehicle fast charging piles are prone to overheating of power devices under high-power operation, causing potential safety hazards. However, the existing cooling strategy adopts a rule-based forced air cooling method, and the cooling fan rotates at a high speed and generates large environmental noise. To protect the thermal safety of core components in the module while optimizing the cooling regulation strategy, an optimal thermal management method for electric vehicle fast charging module based on data-driven model predictive control (MPC) is proposed. This method adopts a data-driven method to construct a prediction model of module temperature distribution based on the long short-term memory neural network, and it combines MPC to control the fan speed, thus optimizing the thermal management strategy for the fast charging module and reducing the fan noise. Through experimental tests, it was verified that this method can effectively reduce the average fan speed by 1 293 rpm and reduce the average noise by 4.99 dB while ensuring that the key components are not overheated, which ensures the thermal safety of core components and the durability of the cooling fan.

The coupling between the boost control and neutral-point voltage balance control of a quasi-Z-source three-level inverter seriously limits its control performance. To solve this problem, a neutral-point voltage balance control strategy based on a virtual space-vector pulse width modulation method is proposed. The neutral-point voltage balance control is realized through a closed-loop control of the DC-bus capacitor voltage, and the low-frequency fluctuations in the neutral-point voltage are eliminated. Meanwhile, a constant shoot-through boost modulation strategy is employed, which avoids the adverse impact on the neutral-point voltage and guarantees an ample boosting capacity of the quasi-Z-source network. Finally, simulation and experimental results verified the validity of the proposed control strategy.

Limited by the switching frequency, the frequency-controlled LLC resonant converter is difficult to achieve a wide output voltage range. To solve this problem, an expandable variable-mode interleaved parallel LLC resonant converter is studied. The secondary-side of this converter adopts a voltage doubling rectifier circuit, which can work in a parallel or series mode according to different switch combinations of two half-bridges on the primary-side, and it can adapt to the wide output voltage range of 1-3N times. A fixed-frequency PWM control method is proposed. In the middle region between the parallel and series modes, the fixed switching frequency is taken as the resonant frequency, and the duty cycle of one bridge arm is changed to realize voltage control. PSIM simulation results show that the wide output voltage range of 1-3N times can be realized by expanding 2N resonator cavities. The experimental results of a 100 W prototype demonstrate that the wide output voltage range of 1-3 times can be achieved with two half-bridges and two resonant cavities, and the effectiveness of the proposed converter and its control strategy was verified.

Since DC bias is one of the main reasons for increases in the vibration and noise of a large-scale transformer, it is essential to fully understand the vibration and noise characteristics of large-scale transformers under DC bias for the evaluation of the operating state of transformers and the reduction of noise and vibration. A 406 MVA EHV large-scale transformer is taken as the research objective, and its vibration and noise characteristics are studied. First, based on the field-circuit coupling finite element method, the no-load operation characteristics under different DC bias currents are simulated and analyzed, and the law of excitation current under different DC biases is analyzed. Second, a multi-physics coupling model of circuit-magnetic field-solid mechanics-pressure acoustics is established, and the effective value of vibration displacement and the time-frequency characteristics of noise signal at different measuring points of the transformer under DC bias are obtained considering the influence of magnetostriction. Third, the sound level is measured at different measuring points around the transformer, and the simulated value is compared with the actual measured value to verify the effectiveness of the proposed calculation method for transformer vibration and noise. Finally, the Hilbert-Huang transform method is used to extract the vibration and noise characteristic quantities of one large-scale transformer under DC bias, and a transformer vibration characteristic recognition method based on the energy ratio of the noise signal intrinsic mode function is proposed. This method can effectively recognize the severity of DC bias of the transformer and accurately grasp its operating state, providing a theoretical basis for timely taking measures to suppress the DC bias.

Aimed at the problems of current unobservable area and zero drift error in the traditional space vector pulse width modulation with single-sensor phase current reconstruction method, an error self-correction complementary non-zero vector pulse width modulation method is proposed. Through the analysis of the DC bus sample principle, the minimum sample time is defined, the complementary non-zero vector is used to replace the zero voltage vector, and the current sampling window is extended, thus eliminating the sector boundary unobservable area. At the same time, the generation mechanism of error amplification effect is revealed, and the zero drift is detected and self-corrected by means of double-sampling complementary non-zero vector, which realizes the compensation for current zero drift reconstruction. Experimental results show that the reconstruction error of the proposed method was lower than 1.26%, and the phase current THD was lower than 6.15%.

In the impedance measurement process, since the inverter impedance varied widely, the magnitude of injection disturbance cannot be evaluated in advance. Therefore, it is necessary to adjust the disturbance energy adaptively. The impedance measurement device of disturbance voltage injected in series is taken as the research object, and an adaptive adjustment strategy of disturbance voltage based on disturbance current feedback is proposed. The magnitude of disturbance voltage is adjusted by detecting the responding disturbance current in real time, thus realizing the adaptive adjustment of disturbance energy. Both the disturbance voltage and responding disturbance current are controlled to be within 10% of the steady-state point of the system under test. The effectiveness of the proposed control strategy was verified by hardware-in-the-loop simulations in real time.

With the increasing penetration rate of renewable energy, carbon emissions are reduced. However, the inherent intermittency and volatility of renewable energy also bring problems such as inertia, security and economy to the power system. The battery energy storage(BES) technology has become one of the important means to solve this problem. Under this background, an autonomous control method for BES system oriented to the active support of grid voltage is proposed based on full-state feedback. First, based on sagging K_v(V_g-v_g) and the virtual capacitor C inertia technology, static power support control and dynamic voltage support control modules are designed, so that the BES system can provide power(static) support and voltage(dynamic) support. Second, the voltage controller and current controller are combined by using the full-state feedback method, which makes the design of the proposed controller more systematic and flexible and reduces the voltage oscillations caused by single-phase ground fault. Third, in order to maintain the stability of state-of-charge(SOC) of BES, a BES SOC controller based on regulatory factors is also designed to further improve the autonomous operation ca-pability of the BES system. Finally, a case study of a 14-node DC system was carried out based on MATLAB and a semi-physical simulation platform, and simulation results verified the effectiveness of the proposed method in the cases of double-support of static power and dynamic voltage and single-line ground fault. With this method, the BES system can be connected to any key node in the grid, and the voltage at the point of common coupling in the grid can be actively supported through the local monitoring of disturbance, which is not affected by disturbance and can be operated and controlled independently. In addition, this method can also prevent the converter from overcurrent during transient low-voltage accidents, so that the autonomous operation capability of BES is realized.

Aimed at the problem that the accuracy of photovoltaic array fault diagnosis based on support vector machine (SVM) is not high and it is easily affected by the kernel function and penalty factor parameters, a photovoltaic array fault diagnosis method based on SVM optimized by the seagull optimization algorithm (SOA) is proposed. The SOA is introduced to optimize the parameters of the SVM model, and an SOA-SVM fault diagnosis model based on the optimal parameters is established. MATLAB software is used to build a photovoltaic array simulation model, and the characteristic parameters under different fault types are extracted and further inputted into the SOA-SVM model for fault diagnosis. Experimental results show that the fault diagnosis accuracy of the SVM model optimized by SOA is significantly improved. Compared with the ABC-SVM and PSO-SVM models, the SOA-SVM model converges faster in the optimization process and has a higher fault diagnosis accuracy.

To solve the difficulty in online life prediction of silicon carbide metal-oxide-semiconductor field-effect transistor (SiC MOSFET) under practical working conditions, a digital implementation method for SiC MOSFET module life prediction based on particle swarm optimization-back propagation (PSO-BP) neural network was proposed. First, the saturation voltage drop of SiC MOSFET was extracted by a saturation voltage drop platform as the temperature-sensitive electric parameter, and a junction temperature prediction scheme based on experimental data was established. Second, a life prediction scheme based on PSO-BP neural network was established by using a power cycling accelerated aging experimental platform to extract the aging characteristic data. Third, the junction temperature prediction scheme and life prediction scheme were transplanted to field programmable gate array to realize the digitization of SiC MOSFET life prediction. Finally, a circuit was designed to verify the proposed method. Experimental results show that the error between the digital junction temperature and real junction temperature was 4.73 ℃, and the percentage of error between the predicted life times and real life times was 4.1%, which proves that the proposed life prediction method is realized digitally and can accurately predict the life times of SiC MOSFET module.

Single-phase charging systems usually face the problem of secondary power pulsation. To solve this problem, a single-phase electric drive reconstructed onboard charger (EDROC) system with low voltage ripple is proposed, in which a Buck/Boost active filter is placed in parallel on the output side of a traditional single-phase EDROC system to absorb or compensate the secondary power pulsation, thereby obviously reducing the output voltage ripple of the charging system. First, the topology, working principle and secondary power pulsation generation mechanism of the single-phase EDROC system are analyzed. Second, a single-phase EDROC system with low voltage ripple is put forward by combing the Buck/Boost active filter. Meanwhile, the topology, working principle, and selection of inductors and capacitors of the Buck/Boost active filter are analyzed in detail. Third, a control strategy for the proposed EDROC system is designed. Finally, a 200 W experimental prototype was designed, and experimental results verified the feasibility of the proposed charger.